225 ACES JOURNAL, VOL. 17, NO.

3, NOVEMBER 2002

COUPLING 2D FINITE ELEMENT MODELS AND CIRCUIT EQUATIONS

USING A BOTTOM-UP METHODOLOGY

E. Gómez1, J. Roger-Folch2, A. Gabaldón1 and A. Molina1

1
Dpto. de Ingeniería Eléctrica. Universidad Politécnica de Cartagena. Campus Muralla del Mar, 30202. Cartagena, Spain
E-mail: emilio.gomez@upct.es, antonio.gabaldon@upct.es, angel.molina@upct.es
2
Dpto. de Ingeniería Eléctrica. ETSII. Universidad Politécnica de Valencia. PO Box 22012, 46071. Valencia, Spain.
E-mail: jroger@die.upv.es

ABSTRACT. The aim of this paper is to present an approach,
able to deal with all possible connections of voltage and
current sources and impedances, combining conductors in
which the skin effect is taken into account and conductors in
which skin effect is neglected. This approach is obtained
using a bottom-up methodology. In this way, the meaning of
terms in the generalized approach is naturally inherited from
some specific problems. This model is presented in a compact
form, preserving sparse, symmetric and positive-definiteness
matrices. The vectors and matrices, computed during the
solution stage, are employed in engineering calculations as
current and inductance computations providing compact
expressions suitable for efficient algorithms. Finally, the
proposed approach, implemented in a FEM software package
developed by the authors, is applied to the study of a three-
phase transformer, supplied with a balanced three-phase
voltage (sinusoidal and nonsinusoidal) and loaded with an
unbalanced three-phase RL impedance. The agreement
between the computed and experimental results shows the
validity of the proposed model and its implementation.
1 INTRODUCTION
The natural excitation for the Finite Element Method (FEM)
in electromagnetism is current (current density). This current
is supplied by an external electric circuit; and, usually, the
value of this current is unknown, as the circuits consist of
passive elements and voltage sources. Therefore, FEM and
electrical equations are coupled by these unknown currents.
There are two different approaches to couple these equations.
In the first approach, called direct coupling, the equations
obtained from FEM and the electrical circuits are integrated in
a single system of equations having the magnetic vector
potential as its solution. In the second one, called indirect
coupling, two different systems of equations are obtained; one
of them takes into account the FEM equations, and the other
one, the circuit equations. In this paper, a method, based on
the direct coupling of FEM and circuit equations is presented.
Many approaches on coupled 2D Finite Element models and
circuit equations have been proposed [1]-[7]. Most of them,
take into account the skin effect for all the conductors or, they
neglect its influence, without evaluating its magnitude over
each conductor. In this paper, a bottom-up methodology is
applied, generalizing the meaning of the studied examples,
instead of a theoretically obtained approach. So, some
common winding connections were studied─one-phase circuit
and three-phase circuits with star and delta connections─,
being presented here the three-phase star connection handling
unbalanced load and voltage conditions. In this paper, the
model is applied to the study of a power transformer, although
it has been applied to other non-static machines as induction
motors, [10].
2 BASIC EQUATIONS
Assuming that the displacement currents are neglected,
Maxwell’s equations using a magnetic vector potential
formulation can be written in 2D as:
¶ æ ¶A(x, y) ö ¶ æ ¶A(x, y) ö
çn ( B ) ÷ + n (B) = - J(x, y) (1)
¶x è ¶x ø ¶y çè ¶y ÷ø
Where A(x,y) is the magnetic vector potential, J(x,y) the
current density (z component) and n the reluctivity that
depends on the magnetic flux density vector (B). In
anisotropic materials, reluctivity depends on |B| and the angle
(q) between B and the rolling direction of the material
n=n(A,q).
The expression of current density depends on how the
conductors are defined. So, the conductors can be modelled as
solid conductors, in which case the skin effect is taken into
account, or stranded conductors, in which case the skin effect
is neglected. Stranded conductors are particularly useful to
model windings in which an individual mesh considering their
relations would exceed in many cases the requirements of
actual computers without providing a significantly greater
accuracy. If all the conductors in the problem are solid
conductors, the following equation is employed, [8] [9]:
¶A
Ñ ´nÑ ´ A = s Eext - s -sn e ´ Ñ ´ A (2)
¶t
The first term on the right side in (2), Eext being the external
electric field and s the electrical conductivity, represents the
current density due to the voltage source. The second term
takes into account the current density due to the temporal

1054-4887 © 2002 ACES

GOMEZ, ROGER-FOLCH, GABALDON, MOLINA: COUPLING 2D FE MODELS AND CIRCUIT EQUATIONS 226

variation of A. The third one represents the current density

iR RlR, LlR
due to the movement, being ve the velocity of the conductor
with respect to B. By using a frame of reference, such that the
relative velocity becomes zero, (2) can be written as, [8]:
eR
dA
Ñ ´nÑ ´ A = s Eext -s (3) VTR
dt VRS eT n
The total current of a conductor can be obtained by integrating
iT RlT, LlT
the right side of (3):
eS
æ dA ö VST i RlS, LlS
I c = òò ç s Eext - s dS (4) S
Sc è
dt ÷ø

In 2D problems (3) becomes: Figure 1. The windings are star connected.

us dA connected (figure 1). They are modelled with FEM, and

-Ñ(nÑA)= s -s (5)
l dt indeed can or can not share the same 2D magnetic circuit. The
total DC resistances, including the resistances corresponding
Where us is the voltage between the terminals of the conductor to the windings and the load, are called RR, RS, RT. In the
and l the length of the conductor (thickness of the considered same way LR, LS, LT take into account the inductance
domain along the z-axis). Equation (5) takes into account the corresponding to the load and the part of leakage flux not
skin effect. considered in the two-dimensional model.
When the conductors are considered stranded, it is supposed From figure1 the following equations can be obtained:
that the current (if) is the same in all the conductors connected
in series (winding); then, their current density can be obtained æ di ö æ di ö
VRS = eR - eS + ç RRiR + LR R ÷ - ç RS iS + LS S ÷ (9)
by J=Nif/S. Where S is the entire section of all the conductors
è dt ø è dt ø
(N turns) integrating the winding, σ the electrical conductivity
and RF the DC resistance. The current can be calculated in 2D æ di ö æ di ö
problems, by considering the voltage uF between the terminals VST = eS - eT + ç RS iS + LS S ÷ - ç RT iT + LT T ÷ (10)
of the winding as the sum of the voltages of its conductors: è dt ø è dt ø

uF l N dA æ di ö æ di ö (11)
VTR = eT - eR + ç RT iT + LT T ÷ - ç RRiR + LR R ÷
if = -
RF RF S òòS dt dS (6)
è dt ø è dt ø

Using (1) and (6):

iR + iS + iT = 0 (12)

N uF s N dA values of eR, eS and eT can be obtained using:

-Ñ( nÑA)=
S RF
-
S òò dt
dS (7)
S æ dA ö
e j = -l N j ç m , j ÷ j = R, S , T (13)
Summarizing (3) and (7) into a single equation for 2D è dt ø
problems:
where Am is the mean magnetic vector potential on j winding,
ì N uF s N dA defined by:
ïï S R - S òò dt
dS Stranded
òò {j ( x, y )} { A ( t )} dS
T
-Ñ(nÑA)= í f S (8)
ï s us - s dA Solid Am, j ( t ) =
Sj
ïî l dt òò dS
Sj (14)

òò b ( x, y ) {j ( x, y )}
3 EXAMPLE. THE WINDINGS ARE STAR T
j dW
CONNECTED
= W
{ A ( t )} = {w} { A ( t )}
T

The authors have solved some complex winding connections. òò dS

In this case, three windings of stranded conductors are star Sj
227 ACES JOURNAL, VOL. 17, NO. 3, NOVEMBER 2002

Being {w}T the phase vector, which provides information Using (20), (21) and the definition of Am in (14) the following
about the topology of each winding, j(x,y) the shape or equation is obtained:
interpolation functions, whose formulas are well-known for
classical elements, and W the entire domain where the integral { J } = Dt [W ][ N ][G ]{u
t+Dt
} a
( t +Dt ) + t
ext
is extended by using the phase functions b j:
-l [W ][ N ][G ][ N ][W ] ({ A } - { A })
T t+Dt t (22)
b
ì 1 If ( x, y ) Î S +

ï
j
- [W ][ N ][G ][G ]{i } a c
t

b j ( x , y ) = í -1 If ( x, y ) Î S - (15)
j
ï 0 Otherwise Where the term [W] is called the phase matrix defined as the
î
matrix containing the phase vectors presented in the problem
Using the Crank-Nicholson time stepping method, the (three in this case).
following equations are obtained:
Using (1) and (22):
t +Dt t

( éë S ){ A
t +Dt
æ 1 ö æ di ö i - i
t
1 æ di ö
ç ÷ + ç1 - ÷ ç ÷ = (16) t +Dt
ùû + l [W ][ N ][Gb ][ N ][W ]
T t+Dt
}=
2 è dt ø è 2 ø è dt ø Dt
Dt [W ][ N ][Ga ]{uext } + l [W ][ N ][Gb ][ N ][W ] (23)
( t +Dt ) + t T
{A }
t

t +Dt t +Dt t
æ 1 ö æ dA ö A - Am
t
1 æ dAm ö
+ ç1 - ÷ ç m ÷ = m (17) - [W ][ N ][Ga ][Gc ]{i t }
ç ÷
2 è dt ø è 2 ø è dt ø Dt
where [St+Dt] is the stiffness matrix. Therefore, equation (23)
Using (9), (10), (11), (12), (13), (16) and (17): models 2D magnetic circuits with a three star connected
ì i t+ü - M S ù ì V t+
windings of stranded conductors, the magnetic vector
RS + V RS ü
Dt Dt
é MT 0 t

potential in the nodes at t+Dt being the unknown values.

R
ï t+Dtï Dt ê ú ï t+Dt ï 2l
í iS ý = 2 ê- M T M R 0 ú í V ST +V tST ý - 2
ï ME ê 0
ï i t+Dt ï t ï ME Besides, current in the windings at t must be computed using
î T þ ë ú t+Dt
- M R M S û î V TR +V TR þ (18) efficient algorithms.
m , R - Am , R
0 ù ì At+ ü
Dt
éM S + M T -M T -M S ù é N R 0 t

ê ú ê ú ï t+Dt ï
M R ú ê 0 N S 0 ú í Am, S - Am, S
t
ê MT MT+M R ý 4 THE GENERALIZED SYSTEM OF
ï ï
ëê M S M S + M R ûú ëê 0 0 N T ûú î Am,T - Atm,T
t+Dt
MR þ EQUATIONS
é M 'R( M S + M T ) M 'S M T - M 'T M S ù ì iR ü
t

1 ê ú ï t ï The example analyzed above can be generalized to different

- M ' M M ' ( M + M ) M ' M í iS ý
ME ê ú
2 R T S T R T R

ëê M 'R M S M 'S M R M 'T ( M S + M R )ûú ï it ï types of connections and conductors (solid and stranded) [11].
î Tþ

æ é t +Dt ù 2s
where:
ç ëS û + [T ] + l [W ][ N ][Gb ][ N ][W ]T ö÷ { At+Dt}=
è D t ø
M j = R j Dt + 2L j j = R, S , T
æ é t ù 2s T ö

M ' j = R j Dt - 2L j j = R, S , T (19) ç - ëS û + [T ] + l [W ][ N ][Gb ][ N ][W ] ÷ { At } (24)

è Dt ø
M E2 = M R M S + M S M T + M T M R + [T ][ sl Q ]{umas ,ext } + Dt [W ][ N ][Ga ]{uext
( t +Dt ) + t ( t +Dt ) + t
}
Equation (18) can be written in a packed form as: - ([W ][ N ][Ga ][Gc ] - [W ][ N ]){i tcab}

{i t+Dt} = Dt [Ga ]{uext( t +Dt )+t } (20)

So, by example, if a specific problem includes windings of
stranded conductors with star connection and other windings
-l [Gb ][ N ] ({ At+m } - { Am}) - [Ga ][Gc ]{i }
Dt t t
with whatever connections, the contribution of the star
connected windings to the terms of (24) is obtained from (23).
The current density can be obtained using:
The matrices in (24) maintain the sparsity, symmetry and
é NR ù positive-definiteness of the traditional FEM. Obviously,
ê 0 0 ú sparsity depends on geometry and the number of nodes since
S
ì J t+
R
Dt
ü ê R ú ì i t+
R
Dt
ü the mean magnetic vector potential introduces connections
ï ï ê NS ú ï t+Dt ï (21) among the nodes of the windings modelled with stranded
{ J t+Dt} = í J t+S Dt ý=ê 0
SS
0 ú í iS ý conductors (14).
ï J t+Dt ï ê ú ï i t+Dt ï
î T þ ê NT ú î
T þ
ê 0 0 ú Defining n, m and p as the number of nodes, and windings
êë ST úû made of stranded and solid conductors respectively, the
GOMEZ, ROGER-FOLCH, GABALDON, MOLINA: COUPLING 2D FE MODELS AND CIRCUIT EQUATIONS 228

meaning and dimensions of the terms in (24) can be expressed
as:
· [S]nxn is the stiffness matrix.
· [T]nxn is called the mass matrix, containing information
Current in windings made of conductors considered as
stranded ones can be calculated using (1), (5) and the
definition of phase vector:
m

about the current distribution. éë S t +Dt ùû { At +Dt }= {I t +Dt }= åN k {w k } ik t +Dt (27)

k=1
· [W]nxm is the phase matrix showing information about the
distribution of windings made of stranded conductors. It is Where m is the number of windings. The current of a winding
composed of the phase vectors defined in (14). (i) can be computed multiplying both terms in (27) by the
· [Ga]mxm, [Gb]mxm and [Gc]mxm take into account how the transposed i-phase vector {wi}T, taking into account that the
windings defined in [W] are connected. In (18) and (20) product of {wi}T and {wk} is null when k is not the same as i:
they are defined to star connections.
{wi } éë S t +Dt ùû { At +Dt }
T
· {At+Dt}n is the vector of unknown nodal values representing
t +Dt
the value of the magnetic vector potential at the current ii = (28)
N i {wi } {wi }
T
time step t+Dt
· {At}n represents the value of the magnetic vector potential
at the previous time step t. This method computes the phase current using as data, the
phase vector, the number of turns, together with the stiffness
· [Q] has the same meaning in solid conductors as [W] in
matrix and the vector of the magnetic vector potential nodal
stranded conductors.
values at the current time step t+Dt. Therefore, all the terms in
{
( t + Dt ) + t
· u mas , ext }
represents the addition of voltages applied to
p (28) are computed during the solution stage of FEM.
the solid conductors at the current time step t+Dt and at the
previous time step t. 5.2 Flux and Electromotive Force Calculations
· {u ext }
( t + Dt ) + t
m
represents the addition of voltages applied to These results can be easily computed using (13) and (14).
the stranded conductors at the current time step t+Dt and at
the previous time step t. 5.3 Inductance Calculations
· {i }t
cab m represents the value of current at the previous time
Inductance has a geometrical meaning when all the materials
step in conductors defined as stranded. exhibit a linear behaviour. In other case, several non-
coincident definitions can be taken into account [12]. Using
5 COMPUTATION OF ENGINEERING
the definition based on flux linkages, self-inductance can be
RESULTS
calculated in windings made of stranded conductors as:
As shown in (23) and (24), currents in windings must be
calculated during the solution stage. Besides, this
Li
t +Dt
=
N i f t +Dt
=
l N i {w i }
T
{A } t +Dt

computational cost must be kept as low as possible. Other

{wi } ùû { A }
t +Dt T
ii éë S t +Dt t +Dt
engineering results such as inductance, magnetic flux and (29)
N i {wi } {wi }
T
electromotive force can be obtained by making use of the
terms defined in (24), computed at each time step avoiding -1
= l N i2 {wi } éë S t +Dt ùû {wi }
T
time consuming calculations in the postprocessing stage.

5.1 Current Calculations In the same way mutual inductance can be computed using the
flux linkage definition:
The winding currents can be calculated using:
l N i{w i} { A }
t +Dt T t +Dt
N f
t +Dt
= i t +Di t =
i f (t)= ò J (t) dW (25) M ij
ij {w j} éë S t +Dt ùû {At +Dt }
T

Wf
(30)
N j {w j} {w j}
T

In solid conductors (25) can be expressed as:

{w j}
-1
= l N i N j {w i} éë S t +Dt ùû
T

æ u ¶A ö
I c = òò ç s s - s dS (26)
¶t ÷ø
As the current is computed in (28), self and mutual
Sc è
l inductances are calculated using the phase vectors and the
stiffness matrix, that can be dependent on time. Stiffness
229 ACES JOURNAL, VOL. 17, NO. 3, NOVEMBER 2002

matrix is affected by the B-H curve when it exhibits a nonsinusoidal three-phase voltage (figure 3), generated with a
nonlinear behaviour. programmable AC power source. These waveforms are
applied to the software package using files obtained from an
In [13] the inductance computation in case of solid conductors oscilloscope. The secondary is loaded with an unbalanced
is stated. three-phase RL impedance. Figure 4 shows the simulated
currents in steady state, and figure 5 shows the computed and
6 EXAMPLES measured currents.

A software package has been developed implementing (24). It i1

1
has been tested with many examples of increasing complexity i2

Current (A)
i3
over transformers, considering several balanced and 0

unbalanced connections in primary and secondary –one phase -1

and three phase transformers–, steady and unsteady states,
sinusoidal and nonsinusoidal voltages and current sources 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Primary phase currents -Time (s)
(balanced and unbalanced); two of these examples are
2 i4
presented below. i5

Current (A)
0 i6

6.1 Steady-state
-2

The model is applied to a 3 kVA three-phase DY transformer 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
(figure 2). The transformer is supplied with a balanced Primary line currents -Time (s)
1 i7
I6 I3 I9
i8
Current (A)

C i9
0
I5 I2 I8
B -1

I4 I1 I7
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
A Secondary phase currents -Time (s)

Load
Figure 4. Primary and secondary computed currents.
Figure 2. The transformer is delta-star connected.

I4 I5 I6
500
2
U
AB 2
Voltage (V)

2
1
0 1
Current (A)

Current (A)

Current (A)

0 0 0
-500
0 0.005 0.01 0.015 0.02 0.025 0.03 -1
-1
Time (s) -2
500 -2
U -2
BC
Voltage (V)

0 0.01 0.02 0 0.01 0.02 0 0.01 0.02

0 Time (s) Time (s) Time (s)
I7 I8 I9

-500 1 1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.5
Time (s) 0.5 0.5
500
Current (A)

Current (A)

Current (A)

U
CA 0 0 0
Voltage (V)

0 -0.5 -0.5
-0.5
-1 -1

-500
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.01 0.02 0 0.01 0.02 0 0.01 0.02
Time (s) Time (s) Time (s) Time (s)

Figure 3. Line-to-line supplied voltages. Figure 5. Computed (solid) and measured (dashed) currents.
GOMEZ, ROGER-FOLCH, GABALDON, MOLINA: COUPLING 2D FE MODELS AND CIRCUIT EQUATIONS 230

20
8 REFERENCES
10
[1] A. Y. Hannalla and D. C. Macdonald, “Numerical
analysis of transient field problems in electrical
Current (A)

-10 machines” Proc. IEE, vol. 123, no. 9, pp. 893-898,

i1 1976.
-20 i2
i3 [2] H. De Gersem, R. Mertens, U. Pahner, K. Hameyer
-30 and R. Belmans, “Coupled field-circuit problem: a
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
FEM/Primary line currents - Time (s) generalized signal flow graph description of the
circuit equations” Eur. Phys. J. AP, vol 1, pp. 247-
20
251, 1998.
10 [3] J. Väänänen, “Circuit theoretical approach to couple
two-dimensional Finite Element Models with
0
Current (A)

external circuit equations” IEEE Trans. Magn., vol.

-10
-20
-30
0 0.01 0.02 0.03 0.04 0.05
LAB/Primary line currents - Time (s)
Figure 6. Computed (FEM) and measured (LAB) currents.
6.2 Non steady-state
The transformer is now YY connected, and loaded with the
same unbalanced three-phase RL impedance. Figure 6 shows
the computed and measured primary currents; the
discrepancies between them are due to the unfavourable
conditions of the high supply voltage and the instant chosen
for the switching-on of the transformer.
7 CONCLUSIONS
A generalized approach for FEM and external circuits
coupling has been presented considering both types of
conductors simultaneously, stranded and solid respectively.
This model has been established using a bottom-up
methodology. The meanings of the terms of the particularized
equation (23) have been generalized to obtain the terms in
general equation (24). Indeed, when a problem containing a
particularized example is selected, the expressions of these are
included in the terms of (24).
Current, magnetic flux, electromotive force and inductance
definitions have been obtained using some terms of (24),
simplifying their notation and speeding-up their computation
during the solution stage rather than during the postprocessing
stage.
Finally, the approach, implemented in a software package
developed by the authors, has been applied to several cases.
The results have shown a good agreement with those
measured in the lab.
32, no. 2, pp. 400-410, 1996.
i1 [4] F. Piriou and A. Razek, “Finite Element analysis in
i2 electromagnetic systems. Accounting for electric
0.06 0.07 0.08 circuits” IEEE Trans. Magn., vol. 29, no. 2, pp.
1669-1675, 1993.
[5] I. A. Tsukerman, A. Konrad, G. Meunier and J. C.
Sabonnadière, “Coupled field-circuit problems:
Trends and accomplishments” IEEE Trans. Magn.,
vol. 29, no. 2, pp. 1701-1704, 1993.
[6] A. Nicolet, F. Delincé, N. Bamos, A. Genon and W.
Legros, “A coupling between electric circuits and 2D
magnetic field modelling” IEEE Trans. Magn., vol.
29, no. 2, pp. 1697-1700, 1993.
[7] J. R. Brauer, B. E. MacNeal and F. Hirtenfelder,
“New constraint technique for 3D finite element
analysis of multiturn windings with attached electric
circuits” IEEE Trans. Magn., vol. 29, no. 6, pp.
2446-2448, 1993.
[8] S. J. Salon, Finite Element analysis of electrical
machines. Kluwer Academic Publishers, 1995.
[9] B. Thidé, Electromagnetic field theory, Upsilon
books, http://www.plasma.uu.se/CDE/Book, 2002.
[10] J. Roger-Folch and V. J. Lázaro, “Contribution to the
transient analysis of induction motors solving the
electrodynamic equations by using the Finite
Element Method”, COMPEL, vol. 14, No. 4, pp. 93-
96, 1995.
[11] J. Roger-Folch, E. Gómez and V. J. Lázaro, “Time-
stepping Finite Element Analysis of Transformers in
2D”, International Symposium on Electromagnetic
Fields in Electrical Engineering, pp. 241-244,
September, Gdansk (Poland), 1997.
[12] D. A. Lowther and P.P. Silvester, Computer-Aided
Design in Magnetics. Springer Verlag, 1986.
[13] J. Roger Folch, V. J. Lázaro and E. Gómez, “Mutual
Inductance Computation using the 2D Finite Element
Method”, International Conference on Electrical
Machines, pp. 1408-1411, Istanbul (Turkey), 1998.
231
Emilio Gómez Lázaro was born in
Albacete, Spain. He received his MSc.
and Ph.D. degrees, from Universidad
Politécnica de Valencia, respectively in
1995 and 2000; both in Electrical
Engineering. He has held teaching and
research positions at Universidad
Politécnica de Valencia and Univer-
sidad Jaume I, Spain. Currently, he is
an Associate Professor of Electrical Engineering at
Universidad Politécnica de Cartagena in Murcia, Spain. His
current research interests include numerical methods for
electromagnetism, large-scale parallel computing and, more
generally, scientific computing and engineering.
Antonio Gabaldón Marín was born in
Murcia, Spain. He received his MSc. in
Electrical Engineering from
Universidad Politécnica de Valencia,
Spain, in 1988, and his Ph.D. degree
from the same University in 1991.
Currently, he is working at Department
of Electrical Engineering of
Universidad Politécnica de Cartagena,
Spain. His research interests include Demand-Side
Management, End-Use Efficiency, Modelling and Distribution
Automation, and numerical methods for electromagnetism.
ACES JOURNAL, VOL. 17, NO. 3, NOVEMBER 2002
José Roger-Folch obtained his MSc. in
Electrical Engineering in 1970 from the
Universidad Politécnica de Cataluña
and his Ph.D. in 1980 from the
Universidad Politécnica de Valencia,
Spain. From 1971 to 1978 he worked in
the Electrical Industry as Project
Engineer. Since 1978, he joined the
Universidad Politécnica de Valencia
and he is currently Professor of
Electrical Installations and Machines. His main research areas
are the Numerical Methods (F.E.M. and others) applied to the
Design and Maintenance of Electrical Machines and
Equipments.
Angel Molina García was born in
Murcia, Spain. He received his MSc. in
Electrical Engineering from Univer-
sidad Politécnica de Valencia, Spain, in
1998. Currently, he is working towards
his Ph.D. at Department of Electrical
Engineering of Universidad Politécnica
de Cartagena, Spain. His research
interests include Demand-Side
Management, Demand-Side Bidding, residential load
modelling, large-scale parallel computing and scientific
computing and engineering.